Is Bilinguals' Color Perception Modulated by Their Active Language? The Case of Lithuanians in Norway

(1)

Is Bilinguals' Color Perception Modulated by Their Active

Language?

The Case of Lithuanians in Norway

November 2021

Master's thesis

Akvile Sinkeviciute

2021Akvile Sinkeviciute NTNU Norwegian University of Science and Technology Faculty of Humanities Department of Language and Literature

(2)

(3)

Is Bilinguals' Color Perception Modulated by Their Active Language?

The Case of Lithuanians in Norway

Akvile Sinkeviciute

Master's Thesis in English Linguistics and Language Acquisition Submission date: November 2021

Supervisor: Mila Dimitrova Vulchanova

Co-supervisor: Natalia Kartushina (University of Oslo) Julien Mayor (University of Oslo) Norwegian University of Science and Technology Department of Language and Literature

(4)

(5)

v

Abstract

Previous research has shown that having a different number of basic color terms across languages modulates speakers’ color perception. Speakers of languages that have two labels for two color tones (e.g., sinyj and goluboj for dark and light blue in Russian) reveal a color category effect: they discriminate between these tones faster when they fall into two categories (one light blue and one dark blue) as compared to when they are from the same category (i.e., both dark blue). On the contrary, speakers of a language that has one basic color term for both of the tones (e.g., blue in English) discriminate the same part of the blue color continuum at the same speed. Basic color terms in Lithuanian and Norwegian divide the blue color spectrum differently. Norwegian, similarly to English, only has one basic color term for blue – blå, while Lithuanian, similar to Russian, has two basic color terms: žydra “light blue” and mėlyna “dark blue”. The first aim of the current study was to examine whether experience with Norwegian can modify the color category effect in Lithuanian speakers of Norwegian (LN). Additionally, the previously reported color category effect in Russian speakers (Winawer et al., 2007) was expected in Lithuanians who live in Lithuania (LL), while Native Norwegians (NN) acted as a control group. LN participants self- reported being proficient in and exposed to their second language, Norwegian, in a variety of daily situations. A speeded color-matching task of blue stimuli that spanned the žydra/mėlyna boundary was employed to investigate color perception’ differences between the three groups. In addition, one more color matching task was performed with verbal interference in Lithuanian for LL, in Norwegian for NN, and in both languages for LN to assess the role of the activated language in color perception. In line with previous research in Russian speakers (Winawer et al., 2007), we found that LL displayed a color category effect, while NN speakers tested on the identical stimuli did not show the same effect.

Moreover, our results revealed that language in which the verbal interference task was performed affected color matching for LN. When the dual task was in Lithuanian, they showed the category effect like LL. However, when the same task was performed in Norwegian, LN did not show any categorical effect. These results demonstrate that (i) color categories in native language affect performance in a color matching task and that (ii) color matching is affected by bilinguals’ activated language, with the color category effect being attenuated when second language is activated.

Keywords: Lithuanian, Norwegian, Linguistic relativism, color perception, categorization, verbal interference, bilingualism, discrimination, reaction time (RT), color category effect (CCE)

(6)

vi

Acknowledgements

Blue are the people here That walk around

Blue like my corvette its in and outside Blue are the words I say

And what I think Blue are the feelings

That live inside me (Jey, Ponte & Lobina, 1999)

First and foremost, I would like to express my gratitude to my supervisors: Prof. Mila Dimitrova Vulchanova, Prof. Julien Mayor & Prof. Natalia Kartushina for the continuous support, inspiration and help throughout the whole research project. I could not imagine working in a better team than with all of you!¹

Mila, thank you for agreeing to be my supervisor, suggesting researching my native language and letting me go on my first mini research trip from Trondheim to Oslo and then to Kaunas (for data collection in Lithuania). I got to experience collecting data at two different host institutions and I enjoyed every moment of it! Julien and Natalia, thank you for adopting a clueless student with open arms and welcoming me to work at your lab at the University of Oslo. Julien, thank you for recommending working on colors, in my now biased opinion, this is one of the most interesting topics I could have chosen! My special thanks goes to Natalia, you have motivated and inspired me in every step of the way from feeding me chocolate and showing me around the lab to teaching me how to design an experiment and explaining the scary data analysis to me.

Besides my supervisors, I would like to thank the amazing people who took time to read my chapters and various drafts of the thesis: Salma, Maciej, Anya, Brooke, and Erlend!

Toma, thank you for all the help in finding Lithuanians in Oslo! Miriam, thank you for translating the experiment instructions for Norwegians! Sofie, thank you for recording yourself for the instructions! I also thank all the other friends that I cannot mention here, because of lack of space. Thank you for listening about my thesis, even if it was not that interesting for you. I have to say that a lot of inspiration to be proactive and maybe a little ambitious came from the conversations at the by Women in Language Science club. Thank you, Evelyn, and Isabella, for hosting the club meetings and introducing, us, young researchers, to the world of academia.

Ačiū visai didelei mano šeimai! Nors pati turbūt niekada nesijausiu užtikrinta dėl savo sprendimų, ačiū jums, kad tiek studijuojant, tiek ir visame kame tiesiog tyliai palaikėte ir niekada neparodėte nė menkiausios abejonės mano pasirinkimais, išklausėte, kai to reikėjo. Mama, ačiū, kad rūpiniesi mano augintinėmis, jei ne tu, Lietuvos palikti taip ir nebūčiau išdrysusi…

Finally, ధన్య వాదాలు to my favorite person, Phani, who loved me on my worst days, encouraged me to not give up on my dreams, and explained the statistics and the math behind it to a total “humanities’ person”!

Above all, this thesis would not be possible without my wonderful participants in Norway and Lithuania. Ačiū/Tusen takk, I hope you had fun with my experiment and please continue contributing to science!

1 Therefore, I use the pronoun “we” thoughout the thesis.

(7)

vii

(8)

viii

List of Figures

Figure 2.1 Linear interpolation from green to blue ...16

Figure 2.2: Linear interpolation from dark blue to light blue ...16

Figure 2.3: The listing of the basic color terms and their universal color categories ...20

Figure 2.4: Stimuli used in Goldstone's study (1998). From “Categorical perception” ....22

Figure 2.5: off-line (A) and on-line (B) effects of language on color perception ...24

Figure 2.7: The design of Winawer et al. (2007) experiment ...26

Figure 2.8: Two-dimensional map for Lithuanian blue colors displaying clustering of tiles for mėlyna and žydra ...29

Figure 3.1: Concepts in L2 users ...38

Figure 5.1: Linear color interpolation of the blue color used in the current study ...45

Figure 5.2: An example of one color matching trial (far comparison) ...45

Figure 6.1: Individual color boundaries for Lithuanian speakers who live in Lithuania ....50

Figure 6.2: Individual color boundaries for native Lithuanian speakers who live in Norway. ...50

Figure 6.3.: Individual color boundaries for native Lithuanian speakers who live in Norway. ...51

Figure 6.4: CCE for LL, and LN in no interference condition ...56

Figure 6.5: The effect of distance among the three language groups (LN, LL, and NN) ..56

Figure 6.6: LL and NN participants under verbal in their respective native languages ....59

Figure 6.7: CCE for LN participants under verbal interference in Lithuanian (left) and no CCE in Norwegian (right) ...59

Figure 6.8: Color distance distortion due to usage of the two BCTs for blue color continuum. Near colors are perceived closer, while far colors are perceived more distant than they are in the equidistant color continuum. ...65

List of Tables

Table 2.1: Color-term usage from 50 Lithuanian participants for tiles and piles ...30

Table 5.1: The color data of the two prototypes for blue color in Lithuanian ...44

Table 5.2: Chroma.js interpolation. Correct lightness + Bezier interpolation ...44

Table 6.1: LME model’s summary for the three groups when the task is displayed without verbal interference. ...54

Table 6.2: LME model (1) summary for Lithuanians (LL) and Norwegians (NN) ...56

Table 6.3: The three-way interaction between language category & distance in the color matching task with interference of LL and NN reported in Mean RTs and SDs ...57

Table 6.4: Model (2) summary for LN (bilinguals) under verbal interference...58

Table 6.5: Mean RTs and SDs in LN participants in the color-matching tasks with verbal interference ...58

(11)

xi

List of Abbreviations

AoA BCT CCE L1 L2 LL LME

LN NN SLA

Age of Acquisition Basic Color Term Color Category Effect First Language Second Language Lithuanians in Lithuania Linear Mixed Effects Lithuanians in Norway Native Norwegians

Second Language Acquisition

(12)

12

The way humans see colors is dependent on how an object absorbs and reflects wavelengths. Only a small electromagnetic field² can be perceived by the human eye, but it is enough to see millions of colors (The Science of How We See Color—And Why We Need Spectrophotometers, 2021). Yet, when we speak, write, and think, we categorize colors based on language-specific labels. Can differences of linguistic color categories’

terminology across languages influence the way we perceive colors? The answer differs between two competing theories in language and color perception research, namely linguistic Relativism and Universalism. Linguistic relativism stems from the classic Whorfian hypothesis that argues that language shapes thought. In contrast, Universalism, based on the research of Berlin & Kay (1969), argues that color perception is universal and is not influenced by the languages we speak. The current work examined languages that differ in their linguistic color categorization: Lithuanian and Norwegian to address whether color perception of Lithuanian-Norwegian bilinguals will be universal despite that their first language (L1) and second language (L2) categorize color differently or will they perceive colors based on the “activated” language. Lithuanian has two basic color terms (BCTs) for blue: žydra “light blue” and mėlyna “dark blue” (Bimler & Uuskula, 2017), while Norwegian has only one - blå “blue”. Essentially, if languages with different linguistic categories affect color perception in monolinguals, then, what impact do languages have in bilinguals’ color perception? We aimed to investigate whether language activation mode in bilinguals can momentarily alter the perception of blue colors towards the monolingual norms of that language. Particularly, we asked whether Lithuanian-Norwegian bilinguals induced into a monolingual Norwegian language activation mode will perceive blue colors like Norwegians or still like Lithuanians?

Many studies that have tested the classic Whorfian hypothesis were conducted on Indo- European languages such as English, Russian, Italian, Spanish, and Greek (Roberson et al., 2005; Gilbert et al., 2006; Winawer et al., 2007; Athanasopoulos et al., 2010; Bimler

& Uuskula, 2014; González-Perilli et al., 2017). The present thesis involved Indo-European languages that, to our knowledge, were not previously researched on color category effects (CCEs): Lithuanian and Norwegian. Lithuanian is one of the two Baltic languages, the other being the neighboring language Latvian. Geographically, Lithuanian is also close to Slavic languages: Russian, Polish, and Belarussian. There are 3.1 million native Lithuanian speakers, most living in the Republic of Lithuania and about 200.000 speakers in other countries (Bilmer & Uuskula, 2017). In 2021, Statistics in Norway reported that there is a total of 40 632 Lithuanians residing in Norway.

Previous research (Winawer et al., 2007) has reported color categorization differences between Russian and English speakers within the blue color continuum. Since Lithuanian is historically, geographically, and linguistically close to Russian, we aim to conceptually replicate Winawer et al. study. Accordingly, the current MPhil thesis concentrates on blue color discrimination differences between Lithuanian and Norwegian languages and seeks to find out whether color labeling/categories differences result in perceiving colors distinctively. Moreover, our design will not only include speakers that have different

2 from about 400 nm to 700 nm ("The Science of How We See Color—And Why We Need Spectrophotometers", 2021)

1 Introduction

(13)

13

linguistic categories for blue color, but also active bilinguals whose two main languages categorize colors differently. We aim to contribute to the debate of whether and to what extent native and non-native languages affect the perception of color categories within the blue color continuum. The CCEs were investigated through behavioral tasks of color matching with and without verbal interference in Norwegian or Lithuanian.

In what follows, the thesis will, first, present the theoretical background on color categorization and perception, cross-linguistic variation, bi/multilingualism, and second language acquisition, then, we will explain the experimental design of the current study and will finalize with an analysis of the study’s experimental data. An experiment was conducted involving three groups of participants: (1) Native Norwegians (NN), (2) Lithuanians in Norway (LN), and (3) Lithuanians in Lithuania (LL). The color matching tasks aimed to determine whether and if so, then to what extent: (i) there are differences of color perception between the three groups, (ii) whether these differences are modulated by a verbal inference task as shown in previous research for the monolingual³ groups (i.e., Winawer et al., 2007)⁴, and (iii) whether the activated language will alter the color perception for bilinguals.

1.1 The Present Study: Research Questions

There were three primary objectives of the current study. Since Lithuanian speakers have two BCTs for blue and Norwegians have only one, we aimed to find out whether this linguistic difference will result in perceiving colors distinctively. Therefore, we aimed to test this possibility in a color matching of blue colored stimuli and to find out whether speakers of Lithuanian and Norwegian show any cross-linguistic differences. The following research question was formed to satisfy the first aim of the study:

1. Do differences in BCTs between Norwegian and Lithuanian impact color perception for native speakers of these languages?

If there were differences in color matching task between these two groups, it would mean that, indeed, one’s native language influences color perception. However, we also anticipated to evaluate whether experience with an L2 can influence color perception, just like the native language does. Since Lithuanians are the second largest immigrant group in Norway, Lithuanian-Norwegian bilinguals are numerous in Oslo and they were tested in addition to speakers who have Norwegian or Lithuanian as their L1 in Norway and Lithuania, respectively. In line with previous research on other languages (Witthoft et al., 2003; Thierry et al., 2009; Athanasopoulos et al., 2011), the level of users’ proficiency and exposure in L2-Norwegian, age of acquisition, and time spent in Norway are likely to influence discrimination of žydra “light blue”. Therefore, we excluded participants who were not proficient in and exposed to Norwegian enough (see Chapters 5 and 6). After selecting Lithuanian immigrants in Norway who were proficient in Norwegian, we aimed to assess whether their L1 color matching habits change because of their experience with the L2-

3 LL and NN participant groups were not monolingual in a sense of knowing only one language. For the sake of simplicity, we refer to them as monolingual throughout the thesis, because they know only one of the target languages (Lithuanian or Norwegian) of this study and live in that language's environment.

4 In Winawer et al. (2007) verbal interference has eliminated CCEs in monolingual speakers of Russian that have the light/dark blue boundary. However, we assumed that for bilinguals the language mode will influence the CCE depending on the activated language’s linguistic categories for blue.

(14)

14

Norwegian. The following research question was formed to satisfy the second aim of the study:

2. Does experience with Norwegian in Lithuanian immigrants lead to matching blue stimuli differently from Lithuanians who live in Lithuania?

Speakers who have two BCTs for blue have been previously reported categorizing colors faster when they belong to different categories (one light blue and one dark blue) and slower when they belong to the same category (i.e., both light blues). Importantly, the reported effect was modulated with a dual-task paradigm. Verbal interference has been found to eliminate (Winawer et al., 2007), diminish or increase (Gonzalez-Perilli et al., 2017) the cross-linguistic effects. Accordingly, we investigated whether verbal interference could disrupt Lithuanians’ performance on the color matching task in the current study. We formed the following research question to meet the third purpose of the study:

3. Does verbal interference attenuate cross-linguistic effects in the color matching tasks for Lithuanians in Lithuania?

If verbal interference can modulate the CCE for Lithuanians who live in their native language’s environment and therefore are not exposed to foreign languages that much, how does the verbal interference affect the bilinguals? We were interested in whether and how the two languages affect color perception on-line, thus we aimed to test LN participants in speeded color matching tasks with verbal interference in both Lithuanian and Norwegian. The fourth research question was formed to discover whether Norwegian and Lithuanian will have different effects for bilingual participants once one of the languages is activated in the mind:

4. Does the language mode (activating Lithuanian or Norwegian), modulate color perception in Lithuanians living in Norway, who have different BCTs for blue in their two languages?

(15)

15

2.1 Language and Thought

The world is full of various sounds, objects, events, and symbols that can be described by words. Human language faculty allows speakers to express their interpretation of the world through a complex linguistic system that stems from cognitive processes shared with other species (Malt and Wolff, 2010). Importantly, this linguistic system enables humans to communicate and provide tools for mental manipulation of knowledge (Gentner, 2003). It is commonly stated that the importance of language lies in communication with others although, according to Chomsky, language is primarily designed to think and interpret thoughts rather than communicate with others (Chomsky, 2015). Thus, even if we do not communicate with others, we are still thinking in a language. Malt and Wolff’s (2010) language-thought interface offered the idea that each culture’s language may portray the world in its unique way due to variations in the lexicons and encoding strategies. One example of this phenomenon that Malt and Wolff presents is that translators sometimes find it hard to transfer “the same” messages from one language into another without somehow adjusting its meaning.

Furthermore, word meanings are thought to be more diverse than general conceptual meanings. Every known culture appears to have a primary color-naming system of some sort. One line of research has focused on the universality of color categorization (Berlin and Kay, 1969), while many others have concentrated on exceptions and variants. Cross- linguistic research shows that one reason for it is that vocabularies partly depend on the community’s physical and cultural environments. Thus, industrialized countries have more words to describe colors than speakers from traditional societies. In the present study, we focused on two languages that name the same part of color space differently (one color name in one language and two color names in another language) and aimed to find out whether this shapes the thought accordingly. Moreover, Malt and Wolff (2010) stated that the way different speakers name such concepts as space, body parts, motion, emotion, mental states, causality, and ordinary household containers differs greatly. According to Malt and Wolff, environment and physical surroundings alone are not a valid explanation to why there are so many of the cross-linguistic differences. The languages of interest in this study, Lithuanian and Norwegian, differ in the way they linguistically categorize blue color continuum even though both Lithuanian and Norway are industrialized countries, and their physical environments are not that different compared to, for example, the environmental difference between Himba speakers from Gabon, Africa⁵ and English speakers from Great Britain.

2.2 Linguistic Relativism

Although human beings can see millions of color shades as recognized by the retina, color perception has been demonstrated to be categorical. It is known that languages have different linguistic categories to describe the color continuum, for instance, blue and green

5 studied by Taylor, Clifford & Franklin (2013)

2 Cross-linguistic Categorization and Color

Perception

(16)

16

colors are linguistically categorizeddifferently in Vietnamese and English (Kay & Maffi, 2008). Importantly, the question of whether the linguistic variation translates into stable differences in color perception when language is not involved is still open and will be discussed in the following paragraphs. Figure 2.1 illustrates a linear color interpolation from blue to green. The English language divides the continuum into two, where one part of the continuum is blue, and another part is green. Individuals might put a line in different places.

Still, most probably, a speaker whose mother tongue is English or another language with one BCT for green and one BCT for blue will partition the continuum somewhere in the middle. However, some languages have one BCT for both green and blue (“grue”), such as Vietnamese.

green blue

“grue”

Figure 2.1 Linear interpolation from green to blue. English and Vietnamese speakers categorize green and blue colors differently: English divided the green/blue continuum into two linguistic categories, while Vietnamese have one linguistic category for both.

The classic Sapir-Whorf hypothesis stands that “we view the world filtered through the semantic categories of our native language” (Regier & Kay, 2009, p. 1). Following the Whorfian hypothesis⁶, the difference in linguistic color categories for green and blue in English and Vietnamese languages (Figure 2.1) would mean that Vietnamese speakers perceive colors differently. Another language, which our research in concerned with, Lithuanian, has two words to describe the blue color, namely mėlyna “dark blue” and žydra

“light blue”, which is different from English or Norwegian that only have one BCT to describe the same color space – blue or blå (Figure 2.2).

mėlyna “dark blue” žydra “light blue”

blå “blue”

Figure 2.2: Linear interpolation from dark blue to light blue. Lithuanians have two distinct linguistic categories (mėlyna and žydra) within the blue color continuum, while Norwegian have only one (blå).

Remarkably, linguistic categories can constrain perception of concepts that refer to time, space, and color. The Dani language of a tribe in New Guinea can illustrate how different color naming patterns can be. The Dani language speakers have only two-color terms: mili which means black, but also refers to cool and dark shades, e.g., blue, green, and black, and mola “white” that also refers to warm and light colors, such as red, yellow, and white

6 Whorf himself never proposed that language should affect color perception (Lupyan et al., 2020)

(17)

17

(Rosch Heider, 1971). The contrast between the Dani language and Indo-European languages, like English, raised the question of whether world languages are as universal as it has been believed. Hence, Sapir and Whorf are the names that stand for the two most defined claims of linguistic relativism. First, if there are structural differences between two languages, then there are also differences in the thinking habits that their respective speakers have. Second, through acquiring one’s native language, one also develops a world view that is not easily changed in later life (Slobin, 1987). The first claim suggests a superficial linguistic difference between languages, like different alphabets, different sounds, word/phrase/sentence structures, and language use should affect our perception and/or cognition. However, this claim does not delve into the matter of how this can happen. The second idea is connected to perceptual differences and is more controversial.

The controversy of the linguistic relativity hypothesis caused debates in linguistics, anthropology, philosophy, and psychology. At times relativism was rejected at all: “The idea that thought is the same thing as language is an example of what can be called a conventional absurdity” (Pinker, 1995, p. 57). Despite the criticism, many researchers are still working on the idea of language carving up the conceptual space in different ways.

2.3 Four Claims by Linguistic Relativists

According to Dedrick (2014), it is widely agreed that there are two versions of the Whorfian hypothesis: “the first is a strong version and claims that language determines or constrains mental operations. The second, a weak version, claims that language influences mental operations” (p. 274). In the studies that support weak relativism, language affected color perception, in both off-line similarity judgments (Davidoff, Davies & Roberson, 1999; 2000) and on-line perceptual discrimination (Winawer et al., 2007; Gilbert et al., 2006;

Drivonikou et al., 2007). Off-line language effects can affect visual perception, but they do not modify the process of perception itself. Besides, the off-line language effect happens after the language processing routines have already been applied. In contrast, the on-line language effect occurs when perception has been affected in that specific moment, for example, by naming the colors (Lupyan et al., 2020). Winawer and the colleagues (2007) tested Russian and English speakers that differ in their linguistic color categories, namely Russians have two BCTs to describe shades of blue – sinyj “dark blue” and goluboj “light blue”, while English uses single BCT – blue. All participants were tested in a forced choice color discrimination task, where Russians displayed a category advantage and English speakers did not. Noteworthy, Russian’s category advantage was removed when they had to discriminate between shades of blue under verbal interference. It was claimed that verbal interference disrupted the use of language and therefore the category advantage in no interference condition occurred because language was involved on-line. Since the present study proposed to conceptually replicate the study on Russian blues by Winawer and the colleagues (2007), we were mainly concerned with the on-line language effects on color categorization. Briscoe (2020) divides the linguistic relativity theory further. He established four claims that are common among relativists:

1. Perceptual grouping, informativeness⁷ and structuring of perceptual color space are universal. However, cultural and pragmatic color categorization are not universally constrained (Roberson et al., 2000;2005; Jameson, 2005). For instance, there is a practical need to communicate the distinction “between edible and non-edible fruits,

7 Informativeness constraint, in color perception, refers to maximization of similarity of colors within category and minimizing similarity of colors between different categories (Gartner, 1974, as cited in Briscoe, 2020).

(18)

18

as well as from the distribution of shades in the natural and social environment.”

(Briscoe, 2020, p. 463)

Although, according to the first claim, salient colors constrain color concept formation, perceptual grouping and informativeness are rather loose restrictions on the construction of color categories (Roberson et al., 2000, as cited in Briscoe, 2020). When an individual is asked to match colors, there is no good way to limit the color space so that the matching of colors would stop at some point. Linguistic relativists do not answer the question on how hue, saturation, and lightness should be weighted in perceptual grouping (Jameson 2005, as cited in Briscoe, 2020). And grouping constraint implies that there has been a prior identification of color space. However, this problem is solved when a color task involves relative similarity: color x is more like color A than like color B. In this model, A and B are fixed and constrained upon assigning shades to categories (Dedrick, 1998, as cited in Briscoe, 2020). This is the model that we have adopted in the present study as well (see Chapter 5 for details). Additionally, the informativeness constraint assures that color systems will help speakers communicate about them but does not specify how many categories a system will contain (Briscoe, 2020).

2. Following Roberson et al. (2000) there are two steps in the conceptual development of colors. First, color categories are formed within the color boundaries that are specific to the color space of the observers, therefore the boundaries are not universal. Second, the best examples or foci of those color categories are taken out.

(Briscoe, 2020)

The second claim has to do with perceptual salience theory which used to be associated with universalism (see Section 2.4). In perceptual salience theory, basic color concepts are formed by positioning boundaries in a color space centered on the “Hering primaries”. The Hering primaries are black, white, unique yellow, unique green, and unique blue and are thought to be salient colors in vision before thought and language (Hering, 1878/1964 in Briscoe, 2020, p. 3). In contrast, the relativists claim that color categories start forming with setting up the boundaries in color space based on specific cultural reasons, environment, and language rather than based on Hering primaries (Briscoe, 2020).

3. Mental representation of colors is language dependent, and BCTs are the primary vehicles of conceptual color categories. Experimental research (Quine, 1973;

Roberson et al., 2000; Davidoff, 2001; Roberson et al., 2005) “would suggest that there are no cognitive color categories that are independent of the terms used to describe them” (Roberson, 2005, p. 66, as cited in Briscoe, 2020).

The third claim is supported by research on patients with color naming impairments and subjects with no impairments who perform a color-matching task under verbal interference. In a verbal interference paradigm, participants are asked to rehearse words (i.e., number combinations) and match colors simultaneously. It was reported that such dual tasks may impair color perception selectively. The current study employed the verbal interference tasks to assess whether language interference will modify the CCE as demonstrated in the previous studies (Winawer et al. 2007, González-Perilli, 2017). Verbal interference is further discussed in Section 2.5. Besides, ordinary subjects who perform color matching with a verbal dual task act similar to aphasic patients in odd-color-out tasks (Lupyan, 2009 in Briscoe, 2020). An individual with impaired color naming cannot sort colored stimuli into groups (Roberson et al., 1999) nor judge which of the three objects differs from the other two in an odd-color-out task (Davidoff & Roberson, 2004). The results of ordinary individuals under verbal interference and color impaired patients form relativist arguments of why the color language may shape our conceptual space.

(19)

19

4. Experience of using a distinct set of linguistic categories, for instance, eleven BCTs in English, can make colors from the same category appear to be more similar to one another, and colors from different categories appear to be more different. In other words, color terminology “distorts perception by stretching perceptual distances at category boundaries”. (Davidoff, 2001, p. 386 in Briscoe, 2020) The fourth claim represents the strong version of linguistic relativism. As mentioned earlier, strong relativism suggests that color perception is categorical, and that language constrains color perception. Briscoe (2020) presented explicit evidence of why color perception is not categorical. For discussion on categorical perception concerning the current thesis, see Section 2.3.

2.4 Universal Color Categories

In the previous section, we discussed Briscoe’s (2020) claims about linguistic relativism which are often contrasted with the universalist approach in literature. Indeed, relativists are primarily seeking differences in the way of cutting up the color space and are less concerned with creating a similarity system in which different languages assign labels to colors. In contrast, Berlin & Kay (1969) proposed a theory on the evolution of Basic Colour Terms (BCTs). Berlin & Kay proposed that color names in the world’s languages are not assigned randomly but systematically. They argue that supporters of linguistic relativism are missing out on not making connections between languages. The latest version of the universalist view is the Universality and Evolution model (Kay, 2015; Kay & Maffi, 1999).

This view suggests that humans have psychologically universal color concepts due to their perceptual salience⁸. The universal color concepts correspond to basic color terms (BCTs) and are simply reflected differently in a wide variety of languages. Hence, according to the universalists, color representation is not dependent on language in thought or perception (Briscoe, 2020, p. 2). Berlin and Kay (1969) changed ideas regarding the Whorfian approach quite drastically by testing 100 languages and finding the system behind the number of color terms a language has. They used standardized color stimuli of 329 color chips from the Munsell color system for gathering the data. The data was collected through two stages: first, informants were asked to list all the BCTs they knew in their native tongue. The data collectors were mostly using language that is native to the informants for instructions and minimizing the use of any other language. Second, every participant had to map both the focal point and the color name boundaries. To avoid taking expressions like blond, salmon-colored, lemon-colored, or blue-green as basic color terms, 4 criteria were established by Berlin and Kay (1969) for a color name to be considered a basic color name:

1. The color name is monolexemic.

2. Its signification is not included in that of any other color term.

3. Its application is not restricted to a narrow class of objects.

4. It must be psychologically salient for informants. This includes a. tendency for a color term to appear in the beginning of color names lists, b. stability of reference among informants and occasions of use, and c. occurrences between idiolects of all participants.

8 In perceptual salience theory, the best examples of the color categories are based on the “Hering primaries”, which are: black, white, unique yellow, unique green, and unique blue salient colors in vision (Hering, 1878/1964, as cited in Briscoe, 2020, p. 3).

(20)

20

According to Berlin and Kay (1969) even though different languages encode in their vocabulary’s different numbers of basic color categories, a total inventory of exactly 11 basic color categories exists from which the 11 or fewer basic color categories of any given language are always drawn. Furthermore, if a language encodes fewer than 11 basic color categories, then there are strict limitations on which categories it may encode (Berlin and Kay, 1969, p. 2). In addition, BCTs evolve through time by incorporating more color terms but still in a constrained sequence. Essentially, the study’s main finding was that there is a hierarchy in which languages express color. If one finds out how many basic terms there are in a language, the hierarchy will tell which colors the language distinguishes.

In languages with two color terms, those will necessarily be black and white (or their equivalents - dark and light). Then if a language has a third color name, it will be red. The following additions to the number of color terms can be from yellow, green, and blue. Next, if a language has a term for brown color, it will appear as the sixth color system. In addition, if there are still more words for colors in a language, those will be purple, pink, orange, and gray.

white and black

red

green yellow

blue brown

orange and/or pink and/or purple and/or gray yellow

green

I II III IV V VI VII

Figure 2.3: The listing of the basic color terms and their universal color categories. Adapted from Basic color terms: Their universality and evolution (p. 4) by B. Berlin & P. Kay, 1991, University of California Press. Copyright 1969 by The Regents of University of California.

Also, Rosch Heider’s (1977) findings on focal versus non-focal colors strengthened the theory of universal color categories. Here, color is considered focal when it occupies a space in one of the color categories (in Figure 2.3) and represents the best example of that category. Rosh-Heider worked with English and Dani informants; the latter only have two basic color names in their language. The results outlined that focal colors are remembered more easily than colors outside of the 11 basic hues for both English and Dani speakers.

Thus, the claim was that the colors might be hardwired into our color vision system and dependent on physiology rather than language. However, in 2000, Roberson et al. failed to replicate Rosch Heider’s findings. The researchers claimed their findings to be supporting linguistic relativism. Moreover, Lucy and Shweder (1979) declare that Rosh-Heider’s work overlooked the fact that color perception affects color memory. Focal colors are easier to look for in a color palette; therefore, they are more perceptually discriminable than the other colors. Thus, the linguistic relativity hypothesis was not discarded by the research of Rosch Heider.

In general, theories of color categorization differ in a variant of imposed language constraints on the formation of color concepts (Briscoe, 2020). The debate of universalist- versus-relativist over language and perception remains. More recently, studies that touch upon the Whorfian hypothesis moved from debating whether language shapes perception to seek observable factors involved in the process of language affecting perception

(21)

21

(Athanasopoulos & Casaponsa, 2020). Nonetheless, it is accepted that the views do not necessarily contradict each other, as in some sense, they are complementary. According to Martinovic et al. (2020), weak relativism “embraces the possibility of new BCCs (basic color categories) and their corresponding basic color terms specific to a given language, beyond the established 11” (p. 1). Moreover, researchers agree that perceptually salient colors do not support color naming patterns and that color categorization effects are relative to a specific language.

2.5 Categorical Color Perception

Categorical perception is warped in that differences between some objects that belong to different categories are highlighted, and differences between objects that fall into the same category are minimized (Goldstone & Hendrickson, 2010). Perceived continuums of similar colors or similar sounds are divided by boundaries and grouped into categories. The transition between two categories in the continuum is often referred to as a category boundary. In the Aristotelian categorization model, boundaries are clear and fixed, whereas, in cognitive linguistics, categories are less clear and can have fuzzy boundaries (Croft et al., 2004). Categorical perception also refers to “faster or more accurate discrimination of stimuli that straddle a category boundary” (Regier & Kay, 2009, p. 439).

In cognitive science, categorical perception presents how our higher-level conceptual and lower-level perceptual systems operate. Therefore, it provides some explanation of how cognition works in general. Typically, perceptual, and conceptual systems are separated by a point at which the information moves only from perceptual to the conceptual system and not the other way around. However, research revealed that categorical effects exemplify information moving both directions. Perceiving things categorically can help our perceptual systems transform sensory signals into internal representations in cognition.

Goldstone & Hendrickson (2010) exemplify the linear signal transition as a staircase function. So, the increasing sensory signal does not affect perception until the signal reaches a certain point where the perception suddenly changes. While the flat portion of the staircase function is happening, different input signals have equivalent effects, but when the input from a different category occurs, the perception changes—treating a range of other stimuli as the same provides a mechanism that justifies equivalence classes.

Successively, equivalence classes provide humans with the formation of symbolic thought, which refers to the mental pictures of symbols (stimuli). The possible explanation why cognitive systems might be built with equivalence classes is that they are rather untrustworthy when it comes to judging superficial similarities or differences (Goldstone &

Hendrickson, 2010). To illustrate, when perceiving two objects, we emphasize the differences between them to belong to different categories and minimize the differences that would put them into the same category. In other words, if we must place both stimuli in the same category, we will try to connect two stimuli, even if they have different apparent forms. However, even when we put two objects into the same category, we may not treat them as the same thing for all purposes (Goldstone & Hendrickson, 2010).

Moreover, entities can be assigned to different categories in different contexts. Thus, if we create a category and name it “pets”, cats and dogs will belong to this category, but if we make a category with the title “pets that meow”, only cats will end up in this category.

According to Goldstone & Henrickson (2010), categorical perception can be learned, and the categories that we have formed or learned influence the way we perceive the objects surrounding us. In the theory of acquired distinctiveness, categories relevant for determining that category’s members become distinctive in general. This theory is based

(22)

22

on Lawrence’s (1949) research with rats. In the series of experiments, rats were trained to discriminate between black and white colors and rough and smooth surfaces. Every time rats choose the stimuli that Lawrence wanted them to choose, they would get a reward.

First, rats were trained to make black/white distinctions. The training phase was conducted as follows: when black shapes were presented, rats were rewarded for a left response, while when white shapes were introduced, the rats were awarded for a right response. In the second training, rats learned better because they already knew the black/white distinction. Therefore, if the stimuli are irrelevant for an earlier training phase, there is a deleterious effect on subsequent discrimination learning. The experiment with rats illustrates the way that the learned categorical perception works. The described impacts are expected in humans and provide structure for categorizing visual discriminations (Lawrence in Goldstone & Hendrickson, 2009).

Furthermore, gradual perceptual warping is thought to be a result of previously learned categories in humans too. Goldstone (1998) aimed to determine whether arbitrary new visual categorizations can be learned and whether they can be learned whether they shape perception. The stimuli that Goldstone used are given in Figure 2.4. First, the portrayed table was given to participants to train them to categorize brightness and size.

Figure 2.4: Stimuli used in Goldstone's study (1998). From “Categorical perception” by Goldstone and Hendrickson, Wiley Interdisciplinary Reviews: Cognitive Science, 1, p. 73.

After the categorization training, participants were given the same/different judgment task.

In the task, either horizontally or vertically neighboring squares were presented, or the same square was repeated twice. Participants were asked to respond whether the two squares were identical on both their size and brightness or differed on either dimension. If dimension was relevant, participants’ same/different judgments together with the entire dimension were more accurate than the same judgments of the participants for whom the dimension was irrelevant and the control group’s participants who were not trained beforehand. Moreover, the subjects were the most accurate when the dimension values were at the boundary between learned categories (i.e., comparing values 2 and 3 on brightness). To conclude, research shows that when participants are taught a new category, it temporally changes their perception. (Goldstone & Barsalou, 1998; Ozgen &

Davies, 2002).

(23)

23

2.6 Categorical Color Perception or Color Category Effect?

The categorical perception was first observed in sound discrimination (auditory). For instance, in English speakers, the sound discrimination was found to be easier for consonant pairs (i.e., /p/ and /b/) that cross the category boundary than for stimuli within the same category (i.e., Flege and Schmidt, 1995). Suppose color perception is like auditory perception. In that case, the shades that look more similar and have the same name should be harder to discriminate than more distant hues with different names, which

“may cause us to perceive color in a more categorical way” (Lupyan et al., 2020, p. 4). For instance, Norwegian speakers do not assign different labels for lighter and darker shades of blue in their language and linguistically treat the whole blue color spectrum as the same.

In contrast, Lithuanian speakers assign two labels to the blue color spectrum and divide it into two language-specific categories. For linguistic relativists, color category boundaries are distortions of continuous color space (Huette & Merced, 2016). As mentioned earlier, linguistic relativists assume that a linguistic difference of color language would translate into a perceptual difference.

However, Briscoe (2020), based on evidence by Roberson & Pak (2009), suggests that discrimination of color is categorical but not perceptual and occurs because of the post- perceptual naming strategy. Moreover, Briscoe (2020) exemplified naming strategy by category effect differences between Tarahumara and English speakers (Kay & Kempton, 1984). Notably, Tarahumara speakers do not have a blue/green boundary, similar to Vietnamese speakers mentioned in Section 2.1. All participants were given triads of colors within the blue/green color space and asked to select a color shade that was the least similar to the other two. English speakers often picked a color from that crossed the blue/green boundary, while Tarahumara speakers did not show a similar effect. Usually, this effect would be referred to as the categorical perception effect. Yet, Kay & Kempton (1984) refer to this effect as a naming strategy that gives advantage for speakers with a greater amount of color labels. In a color trial of their study, when an English speaker saw three color patches, they thought of a strategy that could be used to complete the task.

One of the possible strategies would be applying labels to these colors. Once an English speaker realized that he or she can label the colors to complete the task, he or she noticed that two color patches (A and B) could be called blue, and the third patch (C) could be called green then. C is then selected as the most different patch. Tarahumara speakers do not use this strategy simply because of not having two different BCTs for blue and green (Briscoe, 2020). Moreover, Roberson et al. (2009) claim that the naming strategy effect, like the one found in English speakers vs. Tarahumara speakers, should be called a color category effect (CCE). We adopted this term in the present thesis as well.

Another account of cross-linguistic differences is that speakers of different languages learn to categorize colors based on their long-term experience of speaking the language is proposed by Lupyan et al. (2020). Learning⁹ and using words such as Lithuanian žydra

“light blue” and mėlyna “dark blue” provides a categorization practice that results in the gradual representational separation of the parts of the color spectrum to which the labels are applied. Different linguistic categorization patterns in Norwegian, using the generic term blå “blue”, are expected to produce different discrimination patterns across the color spectrum (Lupyan, 2012). It does not mean that Norwegian participants will not have the capacity to distinguish light and dark blue colors, but, rather, that having two color labels will affect the performance of Lithuanians in terms of speed: they are expected to match

9 in terms of acquired distinctiveness theory.

(24)

24

colors between categories faster than the colors within the same category. Lupyan (2012;

2020) refers to change of color representation over time as an off-line language effect. In Figure 2.5, (a) illustrates the off-line effect, where color patches are represented as steps of equal distance from each other. In contrast, on-line effects of color labels on color representations are portrayed in (b). Here, categorical effects occur by activating a verbal label (or BCT) on-line. The perceived color actives a BCT, which then warps the color representation in the moment (Lupyan et al., 2020)¹⁰. According to Lupyan (2020), the on-line labeling process is covert when color labels are automatically activated by perceptual inputs and can be “further exaggerated by overt labeling such as actively naming a color or reading/hearing a color term” (Lupyan et al., 2020, p. 6).

Figure 2.5: off-line (A) and on-line (B) effects of language on color perception. From

As presented in Figure 2.5, a stronger categorical representation of color patches is more expanded around the category boundary and in the middle of the category. According to Forder & Lupyan (2019), the expansions around the boundary and collapsing of the space within a category result in improved discrimination of atypical hues from more typical ones and in worse discrimination within the same category.

In 2016, Witzel & Gegenfurtner presented categorical facilitation - a weaker version of categorical perception. They tested categorical perception for brown and red, colors that have a boundary which is the least prone to “spurious effects of low-level mechanisms”

(Witzel & Gegenfurtner, 2016, p. 540). First, they determined the boundary for red/brown through a color-naming task, and then they measured it for the just noticeable differences.

The assumption was that if the effect is categorical just noticeable differences would decrease towards the boundary. However, the results did not confirm this assumption.

Second, they measured the performance in reaction times (RTs) and error rates in a speeded discrimination task with color pairs equalized based on observed just noticeable differences. This resulted in slower RTs and error rates for identifying color differences in equal color pairs when the colors crossed the boundary. The results suggest that categories can only facilitate the identification of perceptual differences at the boundary. According to this approach, discrimination performance for colors that cross a category border should be better than for colors that belong to the same category when controlling for low-level sensitivity.

10 In Figure 5, thicker lines denote the labels of “blue” and “green” that are stronger members of those categories than other blues and greens (Lupyan et al., 2016)

(25)

25

2.7 Behavioral Tasks Differences on Color Perception

What makes colors a suitable research domain is that the physical spectrum of light spans continuously, while color perception is often thought to be categorical (Lupyan et al., 2020). To get evidence for the Whorfian hypothesis, first, the two languages that differ in some linguistic features must be identified. Second, these differences need to be mapped onto differences in cognition or perception. Our study has selected Lithuanian and Norwegian languages that have different linguistic categories within the blue color continuum. In the studies of categorization and color perception participants are often asked to either name or match colors. In our study design both color identification and categorization tasks will be employed (see Chapter 5). Color naming tasks usually require participants to choose a color patch from the whole color palette and label it (Berlin & Kay, 1969) or mark color patches within some part of the color space (i.e., green-blue-purple space in Bimler & Uuskula, 2017). The methodology proposed by Berlin and Kay consists of two parts. First, the list task, also known as the term elicitation task, aims to find the most salient BCTs in a language. In this task, participants must make up a list of all color terms that they can recall within a limited time. Second, subjects are engaged in the color- naming task – they are asked to name the colors they saw. In this procedure, subjects were mapping verbal terms with the color stimuli in the color space and establishing potential boundaries of where the color categories begin and end. The responses determined a color naming pattern of a particular language. Furthermore, the most chosen color patches were elicited to be the best examples of BCTs by speakers of different languages. Typically, the best examples BCTs clustered in small regions of the palette, which was the main argument of why the color categories across languages are universal and constrained to a set of 11 color names (Kay et al., 1991; Kay & Maffi, 2000).

Sometimes, a smaller number of color patches are selected by the researchers. In this type of investigation, the participants must name the chosen color patches with distinct BCTs from their language. To analyze color boundaries within the blue color continuum, Bimler

& Uuskula (2017) asked speakers of six languages to sort color patches by similarity and to label the piles afterward. In data analysis, the researchers created an indicator of color term basicness, namely clustering index. The clustering index measured the expansion of light/dark separation and the weight of any category boundary between them within the blue color continuum. Clustering of blue stimuli confirmed “light blue” to be a separate basic color category for 4 out of 6 tested languages. Therefore, “light blue” shall be considered the 12^th basic color category in the Berlin & Kay system for some languages, including Lithuanian.

Forced-choice designs will be referred to as color matching tasks. In a color matching task, the colors are already selected by the researcher; these are typically the colors surrounding the prototypes or the best examples of BCTs known from prior studies of color naming.

Participants are asked to choose between three or two given colors. For instance, in Roberson & Davidoff’s (2000) study, participants were shown a target color chip picked from a blue-green continuum. After the color chip was taken away and two new chips were presented. The participants had to indicate which of the two new chips matched the target chip shown initially. These two new chips were from the same category (both blue or both green) or different categories (one blue and one green. In addition, the two new chips were always normed so that colors within category and colors between categories were the same perceptual distance apart. The results showed that participants were more correct on trials between-category trials than those from within-category trials. Furthermore, Winawer and colleagues (2007) looked for categorical effects in Russian speakers. Russians, unlike English speakers, use BCTs for dark blue (sinyj) and light blue (goluboj). In this study,

(26)

26

individuals saw three colored squares – one on the top and two on the bottom (Figure 2.6).

The aim was to find which colored square of the two bottom ones was identical to the square on the top. Russian speakers were slower (measured by reaction times) to match colors when they belonged to the same category and faster when they belonged to different categories. English-speaking participants did not display the same cross-category effect.

Figure 2.6: The design of Winawer et al. (2007) experiment, where a participant has to decide which of the two bottom colors is identical to the one at the top. From “Russian blues reveal effects of language on color discrimination” by Winawer et al., Proceedings of the national academy of sciences, 104, p. 7781. Copyright 2007 by National Academy of Sciences, U.S.A.

In fact, color discrimination and matching tasks can be modified by manipulating linguistic factors. Several studies have shown (Roberson & Davidoff, 2000; Winawer et al., 2007) that verbal interference tasks can make the cross-linguistic effects vanish. When dual task methodology is employed, participants are instructed to use language to perform a verbal task (e.g., rehearsing digit combinations) and perform the nonverbal perceptual/cognitive task (e.g., categorizing colors) at the same time (Athanasopoulos & Casaponsa, 2020).

In dual tasks, verbal interference disrupts the ability to use color language when matching colors. Color matching tasks together combined with verbal interference tackle the verbal rehearsal system (the phonological loop)¹¹. The phonological loop allows subjects to rehearse stimuli verbally in short-term memory to store these stimuli in long-term memory (Athanasopoulos & Casaponsa, 2020). Crucially, it was claimed the phonological loop can no longer be adopted to process color stimuli that have verbal labels (Athanasopoulos &

Casaponsa, 2020).

In Roberson and Davidoff (2000), subjects were presented with a target color square. After seeing the target as the verbal interference, subjects had to read words for five seconds and then were tested on the two test chips. It was found that the interference condition eliminated the advantage of between-category trials (or the CCE). Winawer et al. (2007) extended Roberson and Davidoff’s study of color memory to on-line color judgments. As mentioned earlier, when language is involved on-line, color perception is being affected in

11 This is based on the well-established psycholinguistic theory of memory by Baddeley (2003), where verbal information is encoded via a phonological loop.

(27)

27

that specific moment. In Winawer et al. 's (2007) study, Russian speakers were slower to match colors when they belonged to the same category and faster when they belonged to different categories. However, when the same task was performed under the verbal interference conditions, matching colors that belonged to the same category no longer took more time than the ones between categories. Thus, the CCE was not only not evident anymore, but also, it was reversed.¹² Another study by Gilbert et al. (2006) employed a verbal interference task that removed the categorical color perception effect found in the right visual field only. Moreover, it is often claimed that the left hemisphere is typically dominant for language (Briscoe, 2020).

Thus, the findings of dual paradigm studies reveal that CCEs are linguistic in their origin.

In other words, the fact that verbal interference tasks eliminate the CCE can justify that this effect is linguistic and not due to cultural or environmental differences between native speakers of different languages (Regier et al., 2010). According to Winawer and Witthoft (2012), the interpretation that the elimination of the CCE is a result of categorical perception is possible, but also is unlikely. So, if CCE occurred because of perception being categorical then the way the color looks was modified only during the moments of color label access. Thus, BCTs are unlikely to influence the early perceptual processes, and the decision process is more plausibly affected by language. For instance, in a matching task, when two distinct colored squares are from the same linguistic category (e.g., they both are light blue), thinking of two distinct BCTs in a language is unlikely to guide a participant towards the correct answer. Yet, if color stimuli in a trial are noticeably having two different BCTs (e.g., one light blue and dark blue), then thinking of these BCTs may speed up memory or the comparison. Following this logic, rehearsing random words or digits interferes with assigning BCTs to the color patches and may eliminate one of the two strategies used for matching the colors. Winawer and Witthoft (2012) conclude that “verbal interference effects are more likely to reflect a role of color terms on decisions, strategy, and memory, rather than perception” (p. 5).

In accordance to Winawer and Wirrholf’s claims (2012), Lupyan (2012) argues that verbal interference shows that the cross-linguistic effect observed without any interference was of language on language. Besides, as it is so easy to eliminate the CCE by the interference tasks, some researchers used it as an argument to claim that the Whorfian hypothesis must be false and superficial (Regier et al., 2010). According to Lupyan (2012), if CCE can be modulated by interference so easily, the question arises whether the effect of language warped perception in the first place? And if it did, why can these effects be “unwarped” so easily? Lupyan (2012) states that the answer has to do with linguistic influences on categorization and perception being superficial. Since CCE was disrupted by verbal but not visual interference, it might be the case that language was affecting the verbal process during the whole task. Ultimately, the cross-linguistic effect may be of language on language and not of language on perception (Gleitman, 2010, as cited in Lupyan, 2012).

The assumption of language influencing language relies on two objectives. First, language is thought to be a “transparent medium” (Gleitman et al., 2004, p. 363, as cited in Lupyan, 2012). Transparent medium occurs when words map onto concepts that are independent words (e.g., Gopnik, 2001; Snedeker and Gleitman, 2004; Gleitman and Papafragou, 2005). Second, it is possible that because verbal and non-verbal processing is strictly separated, the verbal and non-verbal representations are too. In conclusion, cross- linguistic effects on color perception can be interpreted as both fragile and pervasive.

12 According to Briscoe (2020), these findings are consistent with the earlier discussed Kempton and Kay’s name strategy theory (p.16)

Is Bilinguals' Color Perception Modulated by Their Active Language? The Case of Lithuanians in Norway

Is Bilinguals' Color Perception Modulated by Their Active

Language?

The Case of Lithuanians in Norway

Master's thesis

Akvile Sinkeviciute

Is Bilinguals' Color Perception Modulated by Their Active Language?

Akvile Sinkeviciute

Abstract

Acknowledgements

Table of Contents

List of Figures

List of Tables

List of Abbreviations

1 Introduction

1.1 The Present Study: Research Questions

2.1 Language and Thought

2.2 Linguistic Relativism

2 Cross-linguistic Categorization and Color

Perception

2.3 Four Claims by Linguistic Relativists

2.4 Universal Color Categories

2.5 Categorical Color Perception

2.6 Categorical Color Perception or Color Category Effect?

2.7 Behavioral Tasks Differences on Color Perception